Acid-Alkali-Acid Treatment (AAA) of Organic Samples

Most organic materials including sediments, charcoal, wood, and plant remains are chemically pretreated by an acid-alkali-acid (AAA) extraction to remove inorganic carbon and humic substances that may be introduced during post-depositional processes. Briefly, samples are treated with 1% HCl (1 hr, 60°C followed by ca. 10 hr, room temp.) in pre-combusted (450°C, 4 hr) centrifuge glasses followed by repeated washes of the residue with Milli-Q water (Millipore, USA) and extraction with 1% NaOH (4 hr, 60°C) yielding an alkali-soluble fraction and a non-soluble residue (humin). The humin fraction is then washed repeatedly with Milli-Q water, treated again with 1% HCl (ca. 10 hr, room temp.), rinsed with Milli-Q water, until a pH of 5 is achieved, and finally dried at 60°C. If samples dissolve completely in NaOH, i.e. have been transformed into soluble humic substances, we precipitate the humic acid fraction in the alkali solution using concentrated HCl. This fraction can give reliable ages for charcoal samples, because it is often composed of products of post-depositional diagenetic alteration in charcoal (Ascough et al. Reference Ascough, Bird, Meredith, Wood, Snape, Brock, Higham, Large and Apperley2010, Reference Ascough, Bird, Francis and Lebl2011), which are of similar age (Pessenda et al. Reference Pessenda, Gouveia and Aravena2001).

Acid Treatment (A) of Organic Samples

In some cases, we reduce the standard AAA extraction to an acid-only (A) extraction using 1% HCl (1 hr, 60°C followed by ca. 10 hr, room temp.). For carbonate-rich samples it may be the case to repeat this procedure with more concentrated HCl. The A-only extraction is frequently applied to very small or fragile plant fragments in order to avoid sample losses occurring mainly during the extraction with NaOH. In some cases A-only is also used for marine sediments (e.g. Hillenbrand et al. Reference Hillenbrand, Smith, Kuhn, Esper, Gersonde, Larter, Maher, Moreton, Shimmield and Korte2009). Another example is the preparation of soil organic matter in studies of organic carbon cycling in soils (except for dating paleosols). Here we apply only 0.5% HCl (1 hr, 60°C followed by 10 hr, room temp.) The objective is to retain fulvic and humic acids, which are essential components of the soil organic matter (Lehmann and Kleber Reference Lehmann and Kleber2015) in bulk soil samples or soil organic matter components/fractions.

Preparation of Dissolved Organic Carbon

For ¹⁴C analysis of dissolved organic carbon (DOC) in marine, lake or soil water, we first acidify the samples to pH 3–4 with HCl (if not done already in the field). The DOC samples are then freeze-dried and transferred into a tin capsule for EA-GIS-AMS analysis.

Collagen Extraction from Bones

Based on the results of a series of tests (Fülöp et al. Reference Fülöp, Heinze, John and Rethemeyer2013), we use the following extraction method for bone collagen. We included a prescreening of the collagen quality by determining the collagen yield and using elemental analysis. The whole bone should have >0.7% N and a C/N ratio of >4, the collagen content of the bone should be >1%, and the C/N ratio of collagen fraction between 2.6 and 3.9 (van Klinken Reference van Klinken1999; Brock et al. Reference Brock, Wood, Higham, Ditchfield, Bayliss and Ramsey2012). The extraction starts with the cleaning of a piece of bone (3–5 g) with Milli-Q water in an ultrasonic bath (2 × 15 min) and, if necessary, with a sequence of organic solvents including hexane, dichloromethane, and methanol using either Soxhlet or ultrasonic extraction depending on the consistence of the sample. The dried samples are decalcified with 1 M HCl (>3 hr, room temp.), humic substances are removed by adding Milli-Q water whereby the pH is increased to about 3 while the sample is placed in an agitating water bath (24 hr, 60°C). Samples are finally converted into gelatin by hot filtration (60°C) through glass fiber filters followed by freeze-drying. If the collagen quality indicators of the bone or its collagen fraction are not in the typical ranges indicating contamination derived from soil/sediment components (e.g. humic and fulvic acids, proteins, nitrates) or from the degradation of the collagen fraction, we apply ultrafiltration of the collagen fraction. The separation of the larger collagen peptides from possible smaller contaminants (e.g. humic and fulvic acids) is achieved with Sartorius Vivaspin® 15 ultra-filters, with a molecular cutoff value of 30 kDa. The ultra-filters are cleaned prior to usage with Milli-Q water (1 hr, 70°C in an ultrasonic bath), and then centrifuged with Milli-Q water (3x 2300 rpm, 5 min) followed by sonication in 0.01 M HCl for 15 min and centrifugation with Milli-Q water (3x 2300 rpm, 5 min). For the ultrafiltration itself, the collagen fraction is centrifuged (2500 rpm, 15 min) and the collagen peptides are recovered from the top of the solution in the vial and freeze-dried.

Carbonate Pretreatment

For carbonate analysis we use a hydrolysis system, which has been modified after the set up described in Wacker et al. (Reference Wacker, Fülöp, Hajdas, Molnár and Rethemeyer2013), in which carbonates are hydrolyzed and then the CO₂ is transferred to the AGE-2 for graphitization. The hydrolysis system consists of a PAL HTC9-xt (CTC Analytics, Switzerland), the PAL headspace injection tool (HS 1000, CTC Analytics, Switzerland), and a heating block in which up to 21 12-mL Labco® vials can be placed. Prior to analysis, carbonate samples are washed in Milli-Q water in an ultrasonic bath. Then the sample surface is leached using 10–15 mL 1 M H₂SO₄ for 2–10 min depending on the sample size. After washing and drying (at least 4 hr, 60°C), the sample is pulverized and transferred to pre-cleaned (dichloromethane rinse, combustion at 450°C) Labco® vials. They are subsequently sealed with septum caps and flushed with He (grade 4.6, 99.996% purity) for 12 min at 100 mL/min to remove atmospheric CO₂ from the headspace. In order to convert the carbonate into CO₂, 0.5 mL 99% ortho-H₃PO₄ is added and the vials are heated (6 hr, 75°C). The evolving sample CO₂ is flushed with He (60 mL/min, 2 min) to the AGE-2 system while water is retained on a phosphorous pentoxide trap.

Processing of CO₂ Samples Using Molecular Sieve Cartridges

We use stainless steel molecular sieve cartridges (MSC) filled zeolite type 13X (Wotte et al. Reference Wotte, Wordell-Dietrich, Wacker, Don and Rethemeyer2017a) for CO₂ sampling in the field, e.g. using respiration chambers or depth samplers (Wotte et al. Reference Wotte, Wischhöfer, Wacker and Rethemeyer2017b). The CO₂ trapped on the zeolite is release by heating the MSC, which is attached to a vacuum rig, within 15 min to 500°C while flushing with He (grade 4.6, 99.996% purity) at 40 mL min⁻¹ over a water trap (dry ice-ethanol, ca. –80°C) and a liquid nitrogen trap for CO₂ collection in glass tubes as described in detail by Wotte et al. (Reference Wotte, Wischhöfer, Wacker and Rethemeyer2017b).

Compound-Specific ¹⁴C Analysis

We isolate individual lipids (e.g. n-alkanes or n-alkanoic acids) from sediments or soils with a pc-GC device as described in Eglinton et al. (Reference Eglinton, Aluwihare, Bauer, Druffel and McNichol1996). In brief, isolation of individual lipids is achieved by repetitive injection on our pc-GC system (Rethemeyer et al. Reference Rethemeyer, Fülöp, Höfle, Wacker, Heinze, Hajdas, Patt, König, Stapper and Dewald2013) that collects individual compounds based on their specific retention time. The trapping of individual compounds is achieved by condensation of these in glass traps attached to the preparative fraction collector (Gerstel, Germany). The isolated material is then recovered with 1 mL dichloromethane (HPLC grade) and transferred to quartz ampoules for ¹⁴C analysis as gas samples. Finally, the solvent is removed by evaporation. The blank levels of this laborious isolation procedure have been monitored with GC standards of known modern and (close to) fossil ¹⁴C concentration (Rethemeyer et al. Reference Rethemeyer, Fülöp, Höfle, Wacker, Heinze, Hajdas, Patt, König, Stapper and Dewald2013). We routinely determine the extent of extraneous modern and fossil C for each sample set processed with pc-GC and accordingly correct the ¹⁴C contents of the samples.

Preparation for Gas ¹⁴C Analysis

For GIS-AMS analysis, samples are prepared by sealed tube combustion. Briefly, samples are weighed into pre-combusted (900°C, 4 hr) quartz glass tubes together with pre-combusted copper oxide (m_sample:m_CuO = 1:60; Santos and Xu Reference Santos and Xu2017) and silver wool (5 mg) to bind sulfur. The tubes are evacuated and flame sealed using a vacuum rig with a turbo-molecular pump (HiCube 80 Eco, Pfeiffer Vacuum, Germany) and combusted (900°C, 4 hr). The CO₂ evolved is cryogenically purified on a vacuum line and transferred into 4 mm glass ampoules fitting into the GIS system. For EA-GIS-AMS analysis, samples are combusted either in tin boats (4 × 4 × 11 mm) or capsules (0.05 mL), which are previously solvent cleaned in the same way as the tin vessels for the AGE-2 system.

Graphitization

AMS graphite cathodes are produced in our laboratory with an AGE-2 system (Ionplus, Switzerland) coupled with an elemental analyzer (VarioMicroCube, Elementar, Germany) for sample combustion (Wacker et al. Reference Wacker, Němec and Bourquin2010; Rethemeyer et al. Reference Rethemeyer, Fülöp, Höfle, Wacker, Heinze, Hajdas, Patt, König, Stapper and Dewald2013). Depending on sample material and size, different tin containers are used in which the samples are combusted in the EA. For most samples we use tin boats with a size of 4 × 4 × 11 mm (H × W × L). If larger sample volumes are required, we use larger tin boats (8 × 8 × 5 mm) all manufactured by Elementar (Germany). Liquid samples are transferred into tin capsules (0.05 mL, Elementar, Germany) and dried therein. To remove contamination on the tin foil, we routinely clean all tin vessels with organic solvents. The washing includes two rinses with acetone (≥99.9%, supra trace), followed by three rinses with dichloromethane (HPLC grade; Rethemeyer et al. Reference Rethemeyer, Fülöp, Höfle, Wacker, Heinze, Hajdas, Patt, König, Stapper and Dewald2013). After combustion in the EA, the CO₂ sample gas is converted to graphite in the reactors of the AGE using hydrogen over iron (Fe) as catalyst (Alfa-Aesar, iron powder, spherical, <10 micron). The Fe powder is usually weighed into the pre-combusted reactor vials of the AGE-2, evacuated and conditioned at the same day.

Optimization of Graphitization Procedure

So far, we kept the amount of catalyst constant at 4 mg because most samples contain about 1 mg C, i.e. at a Fe:C ratio of 4:1. However, frequently samples are smaller than expected, therefore we tested the dependency of the Fe:C ratio on the background using ¹⁴C-free blank material with 1 mg C, which were graphitized in triplicates with various amounts of Fe as catalyst ranging from 3 to 12 mg Fe (Figure 1). The ¹⁴C concentrations do not vary significantly except for the samples with a Fe:C ratio of 10:1, where the F¹⁴C values are slightly higher. We observe a slight decline of the ion current once the ratio exceeds values of 8:1 in correspondence with Santos et al. (Reference Santos, Moore, Southon, Griffin, Hinger and Zhang2007a, Reference Santos, Southon, Griffin, Beaupre and Druffel2007b), who also recommend not reducing the amount of catalyst for smaller sample sizes, because of more stable beam currents with Fe:C ≥5. Lower amounts of catalyst (Fe:C <4) also do not affect the C yield and quality during the graphitization process (Němec et al. Reference Němec, Wacker and Gäggeler2010). Overall, the results demonstrate that it is not necessary to very accurately weigh the Fe powder because it has a minor effect on data quality.

Figure 1 Blank values (black circles) and ¹²C beam current (open circles) versus Fe:C ratio generated with ¹⁴C-free coal (POC#3; 1 mg C) graphitized with various amounts of Fe as catalyst (3–12 mg Fe).

Long-Term Monitoring of Standard Blanks and Oxalic Acids of Graphite Samples

Figure 2 shows the long-term variability of our normal-sized (650–1000 µg C) process blank, coal (Pocahontas POC#3, Argonne Premium Coal, USA), and the oxalic acid standard (OX-II, NIST SRM 4990C; nominal value: 1.3407 F¹⁴C) in tin boats (4 × 4 × 11 mm) since 2011. We do not use the long-term blank for correction our results but the average of the blanks (3–4) processed with each sample set (usually 12–14) analyzed during one AMS ¹⁴C measurement slot to best account for blank contribution. The average ¹⁴C concentration of this seven-years blank (excluding 6 outliers at the end of 2013 that were due to a dirty ion source) is 0.0012 ± 0.0004 F¹⁴C corresponding to 54,305 ± 2581 yr BP (n = 484; mean size 989 ± 44 µg C; Figure 2a). The monitoring of these standards allows to immediately identify issues with either the graphite production or the AMS measurement itself and to analyze the accuracy and precision (Beverly et al. Reference Beverly, Beaumont, Tauz, Ormsby, von Reden, Santos and Southon2010).

Figure 2 (A) Blank levels (in F¹⁴C) based on measurements of normal-sized ¹⁴C-free coal (POC#3, 650–1000 µg C) graphitized and analyzed since 2011. Results shown are not blank corrected. (B) Results for normal-sized oxalic acid (OX-II) standards (650–1000 µg C) measured since the installation of the laboratory in 2011 are given with blank subtraction.

Additionally, depending on the sample size and sample type, we use different tin containers for combustion and prepare size-matched blanks in the containers used for the samples in order to accordingly correct for any contamination resulting from the tin foil or from sample handling and processing. Samples with extremely low organic carbon contents often require the combustion of very large volumes that do not fit into the normal sized tin boats (4 × 4 × 11 mm). Such samples are combusted in larger tin boats (8 × 8 × 15 mm), which have a slightly higher blank (650–1000 µg C) of an average 0.0016 ± 0.0004 F¹⁴C (n = 145; 51,982 ± 1843 yr BP) for normal sized (650–1000 µg C) samples. Liquid samples are routinely combusted in tin capsules (0.05 mL) that yield average blank values for POC#3 of 0.0017 ± 0.0005 F¹⁴C (51,639 ± 2224 ¹⁴C yr BP, n = 19).

A total of 698 oxalic acid standards were measured during the last seven years with a mean sample size of 994 ± 21 µg C and an average F¹⁴C value of 1.3408 ± 0.068 corresponding to an overall precision of 0.5% for modern material (Figure 2b).

Blank Values for Small Samples

Depending on the sample size we routinely adjust the size of the blank material in order to accordingly correct for background. Figure 3 shows the size dependency of the graphitization blanks for different tin vessels used for EA combustion. The smaller the sample size, the more they are affected by a contamination and its uncertainty. For small sample sizes ranging from 250–650 µg C we still have a reasonable blank of 0.0025 ± 0.0007 F¹⁴C (48,441 ± 2396 yr BP, n = 64) for our standard tin boats and of 0.0030 ± 0.0014 F¹⁴C (47,763 ± 4545 yr BP, n = 12) for the tin capsules. The larger tin boats have a higher blank of on average 0.0043 ± 0.0012 F¹⁴C (44,202 ± 2619 yr BP, n = 9), which can most probably be attributed to impurities of the tin vessel itself, which cannot be removed by solvent washing. We applied the model of constant contamination by fitting the data set of these small blanks with the Monte-Carlo approach. For the tin capsules and smaller tin boats we determined an addition of 0.92 ± 0.05 µg C, while the constant contamination for the larger tin boats is slightly higher (1.42 ± 0.08 µg C).

Figure 3 ¹⁴C concentrations of blank material (coal, POC#3) versus sample size using different tin vessels for EA combustion (circles: tin boats 4 × 4 × 11 mm; squares: tin boats 8 × 8 × 15 mm; triangles: tin capsules 0.05 mL). The curves are based on the model of contamination corresponding to 0.92 µg C modern contamination (usual tin boats and capsules, black line) and 1.42 µg C modern contamination (larger tin boats).

Blanks and Standards of Gas Targets

Blank values and sample size limitations of GIS-AMS and EA-GIS-AMS analysis are discussed in detail elsewhere (Melchert et al. Reference Melchert, Stolz, Dewald, Gierga, Wischhöfer and Rethemeyersubmitted to this issue; Stolz et al. Reference Stolz, Dewald, Altenkirch, Herb, Heinze, Schier, Feuerstein, Müller-Gatermann, Wotte, Rethemeyer and Dunai2017, Reference Stolz, Dewald, Heinze, Altenkirch, Hackenberg, Herb, Müller-Gatermann, Schiffer, Zitzer, Wotte, Rethemeyer and Dunai2019). In summary, both gas injection systems have very similar blank values corresponding to the addition of a constant contamination of about 0.27 ± 0.05 µg C for GIS-AMS and 0.27 ± 0.02 µg C for EA-GIS-AMS introduced during sample processing, combustion, and analysis.

Blank of Molecular Sieve Cartridges

The contamination of CO₂ samples, which are processed with the MSC, by modern C and fossil C is less than 3 μg C and 2 μg C, respectively. The process blank for the entire CO₂ sampling and purification procedure including CO₂ sampling with respiration chambers or depth samplers is ∼0.004 F¹⁴C (44,000 yr BP) and ∼0.003 F¹⁴C (47,000 yr BP), respectively, for sample sizes in the range of 5000 to 7000 µg C. If CO₂ samples sizes are large enough yielding >500 µg C, the CO₂ on the MSC can also be released into the AGE for its conversion to graphite, which produces very low blank values of about 0.0028 ± 0.0002 F¹⁴C (47,000 yr BP). Further details are given in Wotte et al. (Reference Wotte, Wordell-Dietrich, Wacker, Don and Rethemeyer2017a, Reference Wotte, Wischhöfer, Wacker and Rethemeyer2017b).

Compound-Specific Blank

We continuously monitored the contamination added during the isolation procedure of individual lipids using pc-GC and the subsequent handling for GIS-AMS preparation. The entire procedure introduces on average 1.1 ± 0.6 µg C (n = 11) modern contamination (F¹⁴C = 1), which was identified using a ¹⁴C-free standard, and on average 0.7 ± 0.4 µg C (n = 10) of ¹⁴C-free contamination (F¹⁴C = 0) identified with a modern standard. In total, 1.8 ± 0.7 µg C with 0.61 ± 0.41 F¹⁴C is added to each sample during the entire isolation procedure, which is comparable to blank levels for pc-GC reported from other laboratories (e.g. Galy and Eglinton Reference Galy and Eglinton2011; Douglas et al. Reference Douglas, Pagani, Eglinton, Brenner, Hodell, Curtis, Ma and Breckenridge2014; Gierga et al. Reference Gierga, Hajdas, van Raden, Gilli, Wacker, Sturm, Bernasconi and Smittenberg2016). However, we also identified some distinct outlier with contamination levels of up to 4.6 µg C of only modern or only dead contamination. These values result form exceptional circumstances (e.g. cross contamination or individual mistakes). The ¹⁴C concentrations for the isolated modern and ¹⁴C-free standards versus their size in comparison with the ¹⁴C contents of the respective unprocessed standards are shown in Figure 4. The given data set does not show a clear trend of a constant contamination like for the small graphite samples (Figure 3). This might be due to the fact that the individual data points were collected over a long period of over 5 years, while the laboratory equipment changed and different persons treated the samples. We also do not see a relationship between the number of injections, i.e. the duration of sample collection, into the pc-GC and the amount of contamination added to the sample as observed by others (Ziolkowski and Druffel Reference Ziolkowski and Druffel2009). Therefore, we are correcting the ¹⁴C concentrations of the isolated samples for a blank contribution that has been individually determined with each sample set.

Figure 4 ¹⁴C contents of a modern (a) and fossil (b) standard isolated with pc-GC versus sample size with 2-sigma uncertainty. The solid line represents the ¹⁴C content of the unprocessed standard.

In order to reduce the amount of contamination, we tested the influence of different methods of solvent removal after the isolation procedure to the addition of extraneous C to the samples. First, dichloromethane was removed from the isolated lipids in the quartz ampoules by using a stream of N₂. During this procedure 2.6 ± 0.3 µg C of modern and 1.0 ± 0.3 µg C (both n = 2) of fossil contamination was added to the samples. The N₂ used in the lab was only technical grade, but an additional gas filtration unit promised to significantly enhance the purity of the gases, which was obviously not the case. Therefore, solvent removal was established to be routinely performed using rotary evaporation, as contamination levels were significantly lower. With this technique we identified a fossil contamination of 0.4 ± 0.3 µg C (n = 2), while the modern contribution was negligible (0.03 ± 0.05 µg C, n = 2). Our laboratory recently moved into a new building with a supply of high purity N₂ (grade 5.0, 99.999% purity) that most probably will lower blank values, which is currently being tested.